A FastAPI-based REST API for serving GloVe word embeddings based on Wikipedia and Gigaword dataset.

• Get word embeddings for single words
• Batch process multiple words
• Find similar words based on embedding similarity
• 300-dimensional word vectors from Hugging Face model hub
• FastAPI-powered with automatic Swagger documentation
• Automated CI/CD pipeline with GitHub Actions
• Docker support
• Code quality checks (Black, isort, Flake8)
• Comprehensive test suite

No special prerequisites required. The model will be automatically downloaded from Hugging Face model hub on first startup.

1. Clone the repository:

   git clone https://your-repository-url.git
   cd your-repository-name

2. Install dependencies:

   pip install -r requirements.txt

For detailed API documentation, see API Documentation.

• GET / - API information
• GET /health - API health check
• POST /embedding - Get embedding for a single word
• POST /embeddings - Get embeddings for multiple words
• GET /similar/{word} - Get similar words

Format your code:

black .
isort .

Run linting:

flake8 .

Run tests with coverage:

pytest tests/ -v --cov=app --cov-report=term-missing

uvicorn app:app --reload

uvicorn app:app --host 0.0.0.0 --port 8080

• GET / - API information
• POST /embedding - Get embedding for a single word
• POST /embeddings - Get embeddings for multiple words
• GET /similar/{word} - Get similar words
• POST /tokenize - Tokenize text using GPT-3/4 tokenizers
• GET /available-tokenizers - Get a list of available tokenizers

This project is configured for deployment on Replit. Simply import the repository and click Run.

1. Build the Docker image:

   docker build -t glove-api .

2. Run the container:

   docker run -p 8080:8080 glove-api

The project includes a GitHub Actions workflow that:

1. Runs on every push and pull request to main/master branches
2. Checks code formatting with Black
3. Validates imports with isort
4. Runs linting with Flake8
5. Executes the test suite
6. Deploys to Replit (on main/master)
7. Builds and pushes Docker image (on main/master)

Required secrets for deployment:

• REPL_ID: Your Replit project ID
• REPLIT_TOKEN: Your Replit authentication token
• DOCKERHUB_USERNAME: Your Docker Hub username
• DOCKERHUB_TOKEN: Your Docker Hub access token

Pre-trained GloVe vectors based on Wikipedia and Gigaword dataset:

• 6B tokens
• 400K vocab
• 300d vectors
• Uncased

This repository uses Git LFS to handle large files. The following file types are tracked by Git LFS:

• .model - Model files
• .npy - NumPy array files
• .bin - Binary files
• .vec - Vector files
• .gz - Compressed files

When cloning the repository, make sure to have Git LFS installed and run git lfs pull to download the model files.

• GloVe Project
• GloVe Paper
• Git LFS Documentation

MIT